balderdashblog

What is knowledge work?

If knowledge emerges from problem-specific consideration of data and information, then knowledge work is the doing that result in knowledge to emerge. This doing is both the transformation of data and information into novel representations, but also brings

Insurance example

Consider an example from my industry: A young customer with no accident history wants to take out insurance for a BMW in New York. A broker sits down with the customer to capture data about them, the type of vehicle, and the type of cover desired. The broker sends this data to the insurance company1. An underwriter receives this input data, which is known as a submission.

By this point, actuaries have already worked alongside the underwriter to accumulate knowledge about the risk associated with vehicle coverage for various customers and vehicles. The underwriter uses this knowledge to enrich the submission data, a process known as rating. During rating, the accumulated knowledge is conditioned on factors such as the customer's accident history, age, state, and the type of vehicle being insured to better assess the specific risk associated with the submission 2.

After assessing the risks, the underwriter calculates the premium associated with the coverage for the given risk and packages it into a quote. Again, determining the correct premium for risk requires the underwriter to apply their accumulated pricing knowledge to the specific situation. This knowledge is built up over time in collaboration with activities in the finance and marketing departments.

Using the risk and price data points, the underwriter decides whether to return a policy to the broker. This is known as quoting. Upon receiving the quote, the broker reviews it in relation to the customer's needs. This process ultimately results in a decision, which is communicated to the underwriter regarding whether to bind the policy. A record is established indicating that the insurance company now has an obligation to the customer in terms of coverages, and the customer has an obligation to the insurance company in terms of premium payments.

Notes

  • This post is part of a series. See here.

Footnotes and references

Footnotes

  1. Using data, information, and frequently even knowledge interchangeably is reasonable. The frame here is: how actionable are the bits within the context? According to Shannon, bits are the foundational element of information, and information is the emergence of meaning that differentiates one set of bits, a message, from another set of bits, another message. ↩

  2. My word choice here is very deliberate. Bayesian prediction is a statistical method that uses Bayes' theorem to update the probability of an event or outcome as new data or evidence becomes available. It conditions prior knowledge (initial probabilities) on new evidence to create updated knowledge (posterior probabilities). This is the foundation for many computational methods for knowledge inference. ↩

Related tags

emerging technologyartificial intelligencedata and decision scienceinsurance